Automatic identification and visualization of reaction mechanisms contained within direct dynamics simulations

Direct dynamics simulations are employed in many areas of chemistry and biochemistry. When paired with an appropriate underlying ab initio, semi-empirical, or DFT-based potential energy surface and proper sampling of initial conditions, direct dynamics simulations provide an atomic-level view of the reaction dynamics within the system of interest, yielding considerable fundamental insights. Moreover, when a sufficient number of simulations are conducted, they provide a wealth of information regarding overall trends in reactivity. However, they also generate large datasets that often require significant manual interpretation through inspection or developing case-specific analysis techniques. Here, we present an analysis method using a multi-tiered graph theory approach, which automatically highlights the most important mechanistic steps present within an ensemble of direct dynamics simulations. The effectiveness of this approach is demonstrated by examining results from three direct dynamics datasets previously reported for systems relevant to the tandem mass spectrometry community.